Performance Prediction of a Solar District Cooling System in Riyadh, Saudi T Arabia

Home , Central solar heating, District cooling

Energy Conversion and Management 166 (2018) 372–384

Contents lists available at ScienceDirect

Energy Conversion and Management

journal homepage: www.elsevier.com/locate/enconman

Performance prediction of a solar district cooling system in Riyadh, Saudi T Arabia – A case study ⁎ G. Franchini , G. Brumana, A. Perdichizzi

Department of Engineering and Applied Sciences, University of Bergamo, 5 Marconi Street, Dalmine 24044, Italy

ARTICLE INFO ABSTRACT

Keywords: The present paper aims to evaluate the performance of a solar district cooling system in typical Middle East Solar cooling climate conditions. A centralized cooling station is supposed to distribute chilled water for a residential com- District cooling pound through a piping network. Two different solar cooling technologies are compared: two-stage lithium- Parabolic trough bromide absorption chiller (2sABS) driven by Parabolic Trough Collectors (PTCs) vs. single-stage lithium-bro- Absorption chiller mide absorption chiller (1sABS) fed by Evacuated Tube Collectors (ETCs). A computer code has been developed Thermal storage in Trnsys® (the transient simulation software developed by the University of Wisconsin) to simulate on hourly basis the annual operation of the solar cooling system, including building thermal load calculation, thermal losses in pipes and control strategy of the energy storage. A solar fraction of 70% was considered to size the solar field aperture area and the chiller capacity, within a multi-variable optimization process. An auxiliary compression chiller is supposed to cover the peak loads and to be used as backup unit. The two different solar cooling plants exhibit strongly different performance. For each plant configuration, the model determined the optimal size of every component leading to the primary cost minimization. The solar district cooling configuration based on 2sABS and PTCs shows higher performance at Riyadh (KSA) climate conditions and the overall cost is 30% lower than the one of the single-stage absorption chiller plant.

1. Introduction different integration scenarios including cogeneration, trigeneration and district heating/cooling. District energy systems have attracted great interest over the last A further milestone in the development of district energy systems decades thanks to the undeniable environmental and economic benefits has been derived from the integration of energy storages in district and the high level of efficiency and reliability. Boran et al. [1] ex- heating [7–9], as well as in district cooling. The storage can be diurnal amined district heating systems and concluded that they are responsible or seasonal [10]. Powell et al. [11] presented a new methodology to for 33% equivalent CO2 savings compared to conventional gas boilers optimize the operation of the cooling unit coupled with tanks. Gang used for heating, and electricity from the grid for electrical demand. et al. [12] investigated a district cooling system with thermal storage Casisi et al. [2] investigated a distributed cogeneration system with a integrated into a trigeneration plant serving a residential compound in district heating network, applied to a real city center: they demon- a subtropical area. The results of their work show that an appropriate strated that both microturbines and ICEs designed for CHP applications design can lead to economic, environmental and energy savings. are competitive with respect to conventional energy supply. Many pa- The integration of renewable sources into district energy systems is pers focus on improving the system at the network level. The network a way to increase the environmental sustainability, in spite of the im- operation is deeply analyzed in [3,4]: the authors developed a detailed pact on the initial investment [8]. Liew et al. [13] analyzed several numerical model based on a quasi-static approach for the hydraulic options to integrate different energy systems including renewable in behavior and a transient model for the thermal one. Despite most of the different types of building complexes. The exploitation of geothermal existing district energy systems are for heating purpose [5], in recent energy for district heating systems has been deeply investigated [14]. years the development of district cooling systems has grown up due to Heat can be extracted from the ground directly or with the use of heat the increase of the global energy demand for air conditioning. The pumps [15]. growth of district cooling systems is linked to the development and Integrating solar energy into district heating systems is a challenge marketing of absorption chillers: Ameri and Besharati [6] evaluated of recent years [16]. The main issues are the variability of the source,

⁎ Corresponding author. E-mail address: [email protected] (G. Franchini). https://doi.org/10.1016/j.enconman.2018.04.048 Received 22 January 2018; Received in revised form 11 April 2018; Accepted 12 April 2018 Available online 03 May 2018 0196-8904/ © 2018 Elsevier Ltd. All rights reserved. G. Franchini et al. Energy Conversion and Management 166 (2018) 372–384

Nomenclature Ecoll heat collected by the solar ﬁeld (MWh) Erad radiant solar energy (MWh) 1sABS single-stage absorption chiller ETC evacuated tube collector

2sABS two-stage absorption chiller G-value,w window solar factor 2 a0 collector optical efficiency GHI global horizontal irradiance (W/m ) 2 a1 collector 1st order loss coefficient (W/m /K) HX heat exchanger 2 2 a2 collector 2nd order loss coefficient (W/m /K ) ICE internal combustion engine 2 Acoll aperture area of the solar field (m ) PTC parabolic trough collector AHU air handling unit Q̇abs cooling power of absorption chiller (1s or 2s) (kW) 2 ̇ BTI beam tilted irradiance (W/m ) Q aux cooling power of auxiliary chiller (kW) ̇ fi Cap capacity of the absorption chiller (kW) Q coll thermal power collected by solar eld (kW) Abs ̇ CC compression chiller Q rad radiant power (kW) CHP combined heat and power SF solar fraction fl CHW chilled water T mean temperature between inlet and outlet ow rates (°C) Tamb ambient temperature (°C) CostAbs unit cost of the absorption chiller ($/kW) 2 Ttank.bottom temperature at tank bottom (°C) Costcoll unit cost of solar collectors ($/m ) 3 Ttank.top temperature at tank top (°C) Costtank unit cost of the tank ($/m ) 2 CR concentration ratio Uroof roof U-value (W/m /K) 2 CW cooling water U-value,w window U-value (W/m /K) 2 DNI direct normal irradiance (W/m2) Uwall wall U-value (W/m /K) 3 E cooling energy produced by the absorption (1s or 2s) Voltank tank volume (m ) abs η ffi – chiller (MWh) e ciency ( ) θ incidence angle (°) Eaux cooling energy produced by the auxiliary chiller (MWh) thus requiring a thermal energy storage system, and the critical over- powered by parabolic trough solar collectors, with operating tempera- heating during the off-heating season. To avoid these problems keeping tures varying in the range 20–160 °C. high the solar contribution, it is essential to design powerful predictive In the open literature, the potential benefits deriving from the in- models of load management and storage tanks: Daniel Trier [17] has tegration of PTCs and 2sABS are substantially unexplored. Starting from investigated district heating systems with a solar fraction of more than previous researches on the modeling and the analysis of buildings and 70%. Often, the solar thermal systems operating in a district heating solar cooling plants [39,40], this work presents a solar district cooling network are coupled with biomass boilers [18]. Therefore, research is model capable of predicting with accuracy the global performance of focused on improving the efficiency of the production plant [19] and the system and the behavior of each single component in design and off- the network [20,21], by means of detailed computer models and si- design operating conditions. mulation tools [22] and genetic optimization algorithms [23]. Subse- Moreover, the paper deals with the system optimization. It is well quently, the integration of the solar contribution is also extended to the known that an optimal plant configuration is crucial for solar driven cooling systems as a direct result of the development of solar cooling technologies, in order to maximize the overall efficiency and to mini- technologies for individual buildings [24]. mize the investment costs, as well documented by Hang et al. [41]. Only a few articles deal with the integration of solar energy in Among the available optimization methods, the integration of dynamic district cooling systems [25–28]. Marugàn-Cruz et al. analyzed the in- simulations based on the Trnsys software and the Hooke and Jeeves tegration of a high temperature solar technology (solar tower) into a optimization algorithm [42] was proved to be a remarkable solution. A district cooling system to enhance the energy surplus in the warm deep explanation of the Hooke and Jeeves direct search algorithm and season [29]. The integration of concentrated solar power plants with its benefits is reported by Kirgat and Surde in [43]. cooling energy production is investigated also in [30], where the au- The paper firstly introduces the Trnsys models developed for re- thors demonstrated that the combined power and cooling production producing the cooling load of a residential compound, the district for plants operating in island-mode can improve the overall global ef- cooling network and the cooling plant based on two different absorp- ficiency. The abovementioned systems are basically large-scale solar tion chillers: double-stage unit powered by a PTC field and single-stage power plants with heat recovery for cooling production via absorption machine driven by ETCs. A set of dynamic simulations have been per- or adsorption chillers. For smaller sizes, typically the solar field is di- formed using the GenOpt optimization tool in order to maximize the rectly connected to the thermally driven chiller [31–33]. The most thermo-economic performance of the systems. common configuration is based on ETCs coupled with a single-stage LiBr absorption chiller [34,35]: evacuated tube collectors with selective 2. Model description surface exhibit efficiency higher than 65% for a fluid temperature in the range 85–100 °C, typical for the generator of 1sABS. A higher overall 2.1. Building model efficiency is expected from solar cooling systems based on two-stage absorption chillers, whose COP is usually around 1.3–1.4. Such units The case study is based on a residential compound in Saudi Arabia. require a supply temperature higher than 150 °C [36]. For these tem- The current governmental vision encourages for the development of perature levels, the best solar field option is a medium concentration new residential districts based on energy-efficient buildings and solar- collector like a PTC. Although parabolic troughs can exploit only the driven technologies. The compound is supposed to be composed by 96 beam radiation, the reduction of the heat loss area permits to keep high single-family houses. Each building is a two-floor villa with total sur- the efficiency for relatively high fluid temperatures (up to 200 °C). face 575 m2 and global volume 1725 m3. The building architecture re- Mazloumi et al. [37] investigated the operation of PTCs in a small-scale produces the standard residential building style in Saudi Arabia. The 3D solar cooling system: they documented a collector efficiency higher software Google Sketch Up coupled with the plug-in Trnsys3D was used than 68% when coupled to a 1sABS. El Fadar et al. [38] presented a to model the building geometry. The building model allows for evalu- numerical study related to a continuous adsorption refrigeration system ating the shading effects and the radiation on walls and windows. Fig. 1

373 G. Franchini et al. Energy Conversion and Management 166 (2018) 372–384

Fig. 1. Residential compound. shows a rendering of the residential compound. 2.3. Solar field and chiller The geometrical model was imported in a Trnsys deck based on the Multizone Building Type to calculate the cooling load over 1-year The cooling load generated by the compound is assumed to be period on hourly basis. The meteorological conditions were derived covered for a percentage by a solar cooling system. Two different from the Meteonorm international database [44]: a detailed description configurations are analyzed and compared: PTC with 2sABS vs. ETC of the ambient conditions is provided in the Section 3.1. Internal loads with 1sABS. The latter is the typical system of most applications at due to appliances, lights and occupancy were evaluated for each room commercial stage. The former is the advanced configuration proposed and considered in the calculation. The variability of internal loads have in the present work. been taken into account according to the building occupancy. The en- For both plant configurations, the solar field aperture area, the hot velope characteristics comply with the Dubai Green-Building Regula- storage tank volume and the chiller cooling capacity have been opti- tion and Specification (the reference standard for new constructions in mized through the software GenOpt interacting with Trnsys. GenOpt is all the Gulf region) and include all the parameters considered for the a tool for multidimensional optimization of an objective function development of the energy model. The Trnsys model performs a de- computed by a simulation program [47]; the Hooke and Jeeves algo- tailed dynamic simulation able to calculate the cooling load of each rithm is the optimization method selected for the cases investigated in room on hourly basis. The main parameters related to the internal loads the present study. This method searches the optimal solution in a spe- together with the thermal characteristics of the envelope are listed in cifi c range (Constrained Optimization) in the presence of continuous Table 1. and discrete constraints on the operation parameters and on the technical data of the plant components. On the base of annual Trnsys simulations, GenOpt determines the optimal values for the variables, 2.2. District cooling network minimizing the budget cost and, at the same time, complying with the solar fraction target according to the Hooke and Jeeves direct search The district piping network was conceived to meet the cooling de- method. Solar fraction (SF) is the percentage of cooling load covered by mand of the compound made of 96 detached homes. The complex is the solar cooling plant: the residual fraction is supposed satisfied by divided into 4 sub-district of 24 houses fed by different ramifications conventional electric chillers. Typically, solar driven absorption chiller delivering the chilled water to the villas. The overall length of the are sized to fulfill the base load, leaving the peak loads to a backup distribution network is 16.8 km. A conceptual diagram is shown in system. For the present study a SF = 0.70 has been considered. Fig. 2. The solar field is supposed located in a dedicated area close to Fig. 3 shows how the computer codes interact with each other. A the central cooling station. Losses in the piping connecting the solar Trnsys building model provides the cooling load for each villa. The field to the absorption chiller were considered negligible. network model calculates the cumulative cooling energy demand, in- The network is designed for a maximum speed of 2.75 m/s in the cluding pipe losses. The Trnsys model of the solar cooling plant simu- main branches and 2 m/s in the smaller branches. The tube insulation lates the operation over 1-year period and calculates the annual solar thickness is 2 in.; all pipeline parameters are summarized in Table 2 and represent the characteristics of commercial products [45]. Chasapis et al. reported in [18] an optimization of the working parameters of a Table 1 district network: in the present analysis, those optimal values have been Building envelope characteristics. considered. Moreover, such values - including the water velocity - are Comfort & Gains Wall layers & Windows consistent with the best practices for network design suggested by the 2 Engineering Association and the pipe manufacturers [46]. Temperature set point °C 24 Uwall W/m /K 0.57 The control strategy of the heat transfer substations is based on a Relative humidity set point % 50 Wall thickness m 0.27 Air changes Vol/h 0.60 Wall solar absorptance %/100 0.30 mass flow modulation. The nominal temperature difference across the 2 Recuperative HX efficiency % 60 Uroof W/m /K 0.30 heat exchangers is equal to 7 °C. The actual delta T between supply and Infiltration Vol/h 0.4 Roof thickness m 0.27 return line at the pumping station allows for determining the heat loss Lighting (peak) W/m2 19 Roof solar absorptance %/100 0.20 2 along the network. Internal gains (peak) kW 3 U-value,w W/m /K 1.9 Occupancy (average - max) Nr. 4–10 G-value,w %/100 0.621

374 G. Franchini et al. Energy Conversion and Management 166 (2018) 372–384

Fig. 2. District cooling network.

Table 2 Table 3 Pipeline parameters. Budget unit costs.

D (m) Insulation Length Linear Surface per Heat loss coeﬃcient Collectors thickness (m) heat loss unit length (W/m2/K) Evacuated Tube Collectors $/m2 500 (inch) (W/m) (m2/m) Parabolic Trough Collectors $/m2 480

Storage 0.08 2.00 7200 0.7925 0.2513 1.2218 Pressurized Hot Water Tank (2sABS) $/m3 1050 0.10 2.00 4800 0.8839 0.3142 1.0903 Hot Water Tank (1sABS) $/m3 450 0.12 2.00 4800 0.9754 0.3770 1.0026 Chiller 1s Absorption Chiller $/kW 400 2s Absorption Chiller $/kW 560 fraction. GenOpt changes the value of the optimization variables from a minimum to a maximum: the set of parameters is entered to Trnsys, which performs the annual simulation. The optimization parameters (minimum and maximum value, re- The objective function is the budget cost of the solar cooling system search step) are shown in Table 4 with the optimized results. The op- (solar field, tank and chiller). Table 3 reports the unit costs considered timization process is based on the Hooke and Jeeves pattern search for the present investigation. The unit cost of the components was method: starting from an initial variable combination, the algorithm provided by the manufacturers (absorption chillers and compression evaluates the objective function by changing each variable value in chiller) or derived from published reports [48,49]. both directions. The configuration is optimized in a two-step process. The objective function is reported in Eq. (1): the optimization involves The first step searches for the combination meeting the desired solar collector area, tank volume and chiller capacity, whose values are multi- fraction. The second step searches for the combination minimizing the plied by their unit costs (Table 3). A penalty coefficient allows for ne- total cost. glecting the solutions that do not comply with the desired solar fraction.

Fig. 3. Simulation and optimization algorithm.

375 G. Franchini et al. Energy Conversion and Management 166 (2018) 372–384

2 Table 4 TT− amb ()TT− amb η =−aa01 −a2 Optimization parameters. BTI· CR BTI· CR (2) Collectors Area 2sABS 1sABS The thermal efficiency η is defined according to ASHRAE (93/2003) Minimum value m2 8000 12,000 standard and rated by SRCC. The parameters a , a and a are respec- Maximum value m2 16,000 22,000 0 1 2 Step m2 500 500 tively the optical efficiency, the first order loss coefficient and the Optimum Value m2 9035 15,600 second order loss coefficient: their values are reported in Table 5 for Storage PTCs and Table 6 for ETCs. Because of the 1-axis tracking system, the Minimum value m3 300 600 troughs do not collect the DNI, but the beam tilted irradiance, defined 3 Maximum value m 800 1400 as follows: Step m3 50 50 Optimum Value m3 400 995 BTI= DNI·cos( θ) (3) Chiller Minimum value kW 1800 1800 The hot water pumped from the storage is delivered to a two-stage Maximum value kW 3000 4000 lithium-bromide absorption chiller. The model includes the efficiency Step kW 200 200 maps provided by a chiller manufacturer and reported in Fig. 4. The left Optimum Value kW 2315 3250 part of the Fig. 4 shows the capacity reduction (fraction of the nominal Minimum cost function value Mio USD 6.053 9.550 capacity) for different cooling water temperatures and chilled water temperatures. The right part shows the thermal input required by the chiller operating at part load (with different cooling water tempera- Table 5 tures) as a fraction of the design heat input. The optimal cooling ca- PTC-2sABS cooling plant characteristics. pacity determined by GenOpt resulted to be 2315 kW. The chiller op- Solar Field (PTC) 2s Absorption Chiller erates in on-off mode (at the design load) and feeds a cold storage tank (Commercial) of 200 m3. The actual chiller COP is affected by the temperature levels Total aperture area m2 9035 Nominal kW 2315 of the external circuits (hot water, chilled water and cooling water) and Capacitya it is computed and updated at each time step. The main operating a Outlet set point temperature °C 170 Rated COP – 1.39 parameters (in design conditions) and the optimal values computed by CR – 60 Hot source °C 121–175 GenOpt are reported in Table 5. range

Optical efficiency (a0) – 0.7719 Storage The heat rejection system is based on a cooling tower; the control ff 2 3 1st order heat loss coe .(a1) W/m /K 0.1803 Hot tank m 400 system adapts the fan speed to keep the cooling water temperature in volume the range 25–30 °C. 2nd order heat loss coeff.(a )a W/m2/K2 0.0258 Cold tank m3 200 2 2 A proactive control strategy has been implemented. When the volume temperature in the buffer tank located between the chiller and the a Chilled water (in-out) 12–7 °C; hot water (in-out) 175–155 °C; cooling district network rises up to 10 °C, the two-stage absorption chiller is water (in-out) 30–35 °C. switched on until the level of 5 °C is achieved. An auxiliary chiller contributes to cool down the cold storage tank if the cooling power ffi Table 6 provided by the absorption chiller is not su cient to satisfy the load. ETC-1sABS cooling plant characteristics. The auxiliary chiller is designed to cover the peak cooling load in the event of a system failure due to lack of radiation or depletion of thermal Solar Field (ETC) Single-stage Absorption Chiller storage. The control system switches off the absorption chiller to avoid (Commercial) crystallization problems when the hot water from the solar field and the Total aperture area m2 15,600 Nominal kW 3250 cooling circuit reach the lower (121 °C) and upper (35 °C) temperature a Capacity limits respectively. – Outlet set point temperature °C 94 Rated 0.723 fi COPa The Trnsys deck for the PTC-2sABS plant con guration is shown in CR – 1 Hot source °C 65–95 Fig. 5. Some ‘types’ (this is the name of the Trnsys routines) are from the range software database and TESS libraries, while other components are ffi – Optical e ciency (a0) 0.718 Storages based on proprietary models, developed by the authors. The type ff 2 3 1st order heat loss coe .(a1) W/m /K 0.974 Hot tank m 995 “ ” fi volume Meteo reads the le of the Meteonorm database, providing solar ra- 2 2 3 2nd order heat loss coeff.(a2)a2 W/m /K 0.005 Cold tank m 200 diation on tilted surfaces, ambient temperature and relative humidity. volume The solar loop, including the collector field and the pressurized hot water tank, is highlighted with the bold red line. The thermal storage is a – – Chilled water (in-out) 12 7 °C; hot water (in-out) 90 80 °C; cooling water linked (orange line) to the absorption chiller, which supplies chilled (in-out) 30–35 °C. water to the cold tank (light blue line). The type “ThermalLoad” reads an external file containing the required cooling load on hourly basis. fmin =+ACol·· Cost Col CapAbs CostAbs + Vol tank · Cost tank + Penalty(0.70SF < ) The green line depicts the cooling circuit connected to the evaporative (1) cooling tower.

2.3.2. ETC-1sABS 2.3.1. PTC-2sABS As mentioned before, the most common solar cooling configuration In the present solar cooling plant configuration the solar field is is based on a single-stage lithium-bromide absorption chiller driven by based on Parabolic Trough Collectors supplying pressurized water to a Evacuated Tube Collectors. The cooling plant configuration is similar to hot storage tank. A variable speed pump regulates the mass flow rate to the one described in the previous paragraph, but with different settings keep the temperature level at the set point (170 °C). The PTC field is and different sizes. Table 6 reports the optimized parameters related to Nord-South oriented and equipped with 1-axis tracking device. The the solar field and the chiller, while Fig. 6 shows the 1sABS perfor- field efficiency is computed according with the quadratic Eq. (2); the mance curves for different inlet cooling water (CW) temperatures. loss coefficients are kept from the data sheet of a commercial trough It can be noted that all the optimized variables (aperture area, and are reported in Table 5. chiller capacity, hot tank volume) exhibit higher values, compared to

376 G. Franchini et al. Energy Conversion and Management 166 (2018) 372–384

Fig. 4. Two-stage absorption chiller performance maps.

Fig. 5. Trnsys deck of the solar cooling plant conﬁguration PTC-2sABS.

Fig. 6. Single-stage absorption chiller performance maps.

377 G. Franchini et al. Energy Conversion and Management 166 (2018) 372–384

Fig. 7. Ambient temperature and relative humidity.

Fig. 8. Global horizontal irradiance vs. direct normal irradiance.

Fig. 9. Building cooling load.

Table 7 although ETCs can exploit the global incident radiation (including the Building cooling load. diﬀuse component).

Peak Load Annual Load 3. Case study and simulation results Sensible Load kW 35.00 Sensible Load kWh 115,338 Latent Load kW 16.84 Latent Load kWh 41,223 Total Load kW 47.31 Total Load kWh 156,561 3.1. Building and compound results

The compound is supposed to be located in the Riyadh area (Saudi the configuration PTC-2sABS. Because of the lower efficiency of the Arabia). In this region, the climate is warm and dry, with humidity single-stage chiller, a larger cooling capacity with a larger storage tank levels very low all year long, as shown in Fig. 7. The ambient tem- are required to meet the cooling demand. Moreover, the collector perature shows both daily and seasonal remarkable excursions between aperture area is significantly higher than the PTC case (+72.5%), maximum and minimum. The air moisture is very low: in the summer

378 G. Franchini et al. Energy Conversion and Management 166 (2018) 372–384

Fig. 10. Cumulated cooling load.

Fig. 11. District network simulation results.

Fig. 12. Water speed in different branches. period, the relative humidity fluctuates around 20%. This leads to a basis over 1-year period and the results are reported in Fig. 9, where the slight diffusion of the available solar radiation allowing concentrated monthly cooling demand is superimposed to the instantaneous cooling solar devices to operate efficiently. In Fig. 8 GHI and DNI are compared: load. The bars show the cooling load variation during the seasons, on annual basis they are respectively 2217 kWh/m2 and 2296 kWh/m2. whilst the fluctuations show the daily trend influenced by solar radia- Starting from the model presented in the Section 2.1, a dynamic tion and ambient temperature. Peak and annual values are summarized simulation of the building thermal behavior was carried out on hourly in Table 7. The annual building cooling load resulted 156,561 kWh with

379 G. Franchini et al. Energy Conversion and Management 166 (2018) 372–384

Table 8 annual demand is 36,381 kWh. The total electricity consumption for the Cooling plant performance. district network pumping stations is 145,525 kWh.

PTC + 2sABS ETC + ABS 3.3. Cooling plant Cooling load MWh 16,265 Cooling load MWh 16,265 Solar energy (based on DNI) MWh 20,741 Solar energy (based MWh 34,590 The annual yield of the two plant configurations is shown in on GHI) PTC Collected heat MWh 12,612 ETC Collected heat MWh 20,432 Table 8. The solar fraction is 0.70 for both cases. In spite of a similar 2s Absorption chiller MWh 11,438 Absorption chiller MWh 11,402 absorption chiller cooling production, the required collected heat is production production significantly higher for ETCs because of the lower efficiency of the Auxiliary chiller production MWh 4774 Auxiliary chiller MWh 4958 single-stage chiller (0.78 vs. 1.36). production Solar Fraction 0.7 Solar Fraction 0.7 Fig. 13 shows the monthly simulation results for the PTC-2sABS case. The difference between the solar energy based on DNI (yellow1 bars) and the collected heat (red bars) indicates the efficiency of the

Fig. 13. PTC + 2sABS monthly results. a peak load of 47.31 kW. The sensible part of the cooling load covers solar field, including the tracking system. The cyan and purple bars about two-thirds of the total load; the low impact of the latent load is show the cooling production of absorption and electric chiller respec- due to the dry climate. tively: the contribution of the auxiliary cooler is higher than 35% from July to September. 3.2. District cooling network Fig. 14 shows in detail the simulation results related to a 72-h period in summer (left) and winter (right), when the cumulative cooling The cooling load was assumed similar for all 96 villas; nevertheless, load is around 2200 kW and 200 kW, respectively. Looking at the solar because of the non-simultaneity of the peak loads, the demand of the field operation, in the summer mornings PTCs collect energy to charge whole compound was shaved: the resulting cumulative load is reported the hot tank and to feed the absorption chiller. In the afternoon the in Fig. 10 with the duration curve. The non-simultaneity of the loads is 2sABS shows an intermittent operation, because the hot tank control considered with a normal distribution that redistributes 10% of the system switches off the pump when the upper temperature limit is compound cooling load in the previous and subsequent hours. achieved (as indicated by the bottom tank temperature reaching the The simulation of the district cooling network was carried out over a level at the top). Furthermore, the chart shows the important role of the 1-year period with a time-step of 0.125 h to ensure a high resolution of hot storage: the absorption chiller operates also nighttime, even if at the results. The computed thermal losses account for approximately partial load because of the decreasing hot water temperature. The 8.2%. Fig. 11 shows a detail of 3-day simulation results: the reported auxiliary chiller switches on for a very short time when the temperature curves indicate the cumulative cooling load corresponding to the 96 in the cold storage tank exceeds the upper limit (10 °C). villas (with peak shaving due to the non-simultaneity) and the dis- During the winter days, the absorption chiller switches on only in- tributed cooling power including piping losses. termittently, due to the very low cooling demand, and the auxiliary The water velocity in the branches of the network with different chiller never operates. diameter is shown in Fig. 12. As previously stated, the flow rate is The simulation results of the single-stage absorption chiller powered regulated in order to keep constant the temperature levels in the net- by the ETC field show a different behavior (Figs. 15 and 16). The work. It can be seen that the maximum speed is lower than 2.8 m/s also amount of solar energy (yellow bars in Fig. 15, based on GHI) exploited in the peak load periods. as source for the district cooling system is 67% higher than the PTC Major losses have been evaluated according to the Darcy equation case, because of the lower collector and chiller efficiencies. In the 3-day using a quasi-static approach; only linear losses were considered. For summer period reported in Fig. 16, the single-stage absorption chiller each time-step, the losses were computed for each branch of the net- operates only daytime, whilst during the night the auxiliary unit must work considering the actual water velocity and the pipe characteristics. Starting from the energy loss calculation, the power of the pumps in the district cooling network was assessed. For each sub-district deli- 1 For interpretation of color in Figs. 13 and 15, the reader is referred to the web version vering chilled water to 24 buildings, the peak power is 20 kW and the of this article.

380 G. Franchini et al. Energy Conversion and Management 166 (2018) 372–384

Fig. 14. (PTC + 2sABS) 72-h simulation results: summer (left) and winter (right).

Fig. 15. ETC + ABS monthly results.

Fig. 16. ETC + ABS 3 day result: summer (left) and winter (right). switch on to feed the district network, because the hot water tem- volume calculated in the optimization procedure. The 1sABS has a perature in the tank falls under the minimum level (70 °C). The running bigger capacity (3250 vs. 2315 kW) and a lower rated COP (0.723 vs. time of the chillers is inﬂuenced by the cooling capacity and the storage 1.39) if compared to the 2sABS. This leads to a higher hot water

381 G. Franchini et al. Energy Conversion and Management 166 (2018) 372–384

Fig. 17. Compression chiller performance maps.

Table 9 Table 11

Water-cooled compression chiller technical data. Primary energy consumption and CO2 emissions.

Compression chiller Primary energy consumption Annual CO2 emission

Nominal Capacitya kW 1000 CC TOE 724.33 CC tonnes/year 2028 Rated COPa – 5.65 1sABS TOE 220.74 1sABS tonnes/year 618 Chilled water kg/h 170,000 2sABS TOE 212.45 2sABS tonnes/year 594 Cooling water kg/h 203,000

a Chilled water (in-out) 12–7 °C; cooling water (in-out) 30–35 °C. Table 12 Budget cost estimation (solar cooling plant). Table 10 PTC + 2sABS ETC + 1sABS Electric energy consumption. – fi – fi Peak Load Annual Load PTC solar eld Mio USD 4.337 ETC solar eld Mio USD 7.800 Hot storage Mio USD 0.420 Hot storage Mio USD 0.450 CC kW 790 CC GWh 2.881 2s Absorption Mio USD 1.296 1s Absorption Mio USD 1.300 1sABS kW 795 1sABS GWh 0.878 Chiller Chiller 2sABS kW 689 2sABS GWh 0.845 Auxiliary Chiller Mio USD 0.480 Auxiliary Chiller Mio USD 0.480 Cold storage Mio USD 0.090 Cold storage Mio USD 0.090 Cooling Tower Mio USD 0.100 Cooling Tower Mio USD 0.100 TOTAL COST Mio USD 6.723 TOTAL COST Mio USD 10.220 consumption, causing a quicker depletion of the hot tank in spite of the larger capacity (995 vs. 400 m3). Moreover, the chiller running time is also related to the performance maps. The 1sABS has a narrower range load. – – of driving temperatures (65 95 °C vs. 121 175 °C): the wider tem- Just to compare the investigated solar district cooling configura- perature dead band of the 2sABS allows for a longer running time (the tions and a conventional one, a further district cooling system with shut down for minimum temperature in the storage occurs later on). identical network and a centralized cooling station based on compres- In winter, the 1sABS switches-on a couple of times per day, and the sion chillers (CC) has been modeled. The cooling station is powered by heat collected by ETCs in the morning is enough to satisfy the daily electricity imported from the grid and includes several chillers with a

Fig. 18. Monthly electric energy consumption.

382 G. Franchini et al. Energy Conversion and Management 166 (2018) 372–384

Table 13 5. Conclusion Budget cost estimation (district network). fi DISTRICT NETWORK In this work two solar district cooling systems including solar eld, cooling plant, district network and building load were modeled and Infrastructure Mio USD 0.600 simulated for Riyadh climate conditions. The cooling load of a re- Branch Mio USD 0.347 sidential compound of 96 single-family detached homes was evaluated Heat Interface Unit & Heat Meter Mio USD 0.254 TOTAL COST Mio USD 1.201 and supposed to be covered by a centralized cooling station with a district cooling system supplying chilled water. The two investigated cooling plants are based respectively on a two- design capacity of 1 MW each. The compression chiller’s Trnsys model stage absorption chiller driven by Parabolic Trough Collectors (PTCs) is an empirical model similar to the one of the absorption chillers: the and a single-stage absorption chiller driven by Evacuated Tube COP evaluation is based on the operating maps provided by the man- Collectors (ETCs). An optimization procedure has been developed to ufacturer and presented in Fig. 17, while the main parameters of the determine the size of all main components assuring the cost mini- compression chiller are shown in Table 9. This CC model was also used mization and an annual solar fraction of 0.7. Transient simulations over to simulate the auxiliary chiller in the 1sABS and 2sABS cases. 1-year period have been carried out on the two investigated plant Table 10 shows the electricity consumption for the three district configurations to evaluate the energy performance under variable op- cooling plants. The electric consumption of the systems based on ab- erating conditions. This simulation procedure allowed to predict the sorption chillers includes the auxiliary chiller, the electric equipment annual operation of the solar cooling systems with high level of accu- and the pumping stations. The three configurations show a similar peak racy. Moreover, the optimization based on the unit costs of the com- load (in the 1sABS and 2sABS cases the peak power consumption is due ponents available on the market permitted a comparison both in terms to the auxiliary chiller), whilst the annual power consumption is about of efficiency and investment costs. The plant configuration based on 70% lower for the solar driven systems (according to the SF). The PTCs and 2sABS resulted to be significantly more cost effective (−30% monthly electricity consumption for the three cooling systems is shown of primary costs, including the district network) than the single stage in Fig. 18: in the winter months, the solar cooling systems do not re- absorption chiller solution for the considered location (Riyadh). This quire the auxiliary chiller operation. result is due to the higher efficiency of the two-stage absorption chiller

Starting from the electrical consumption, the amount of CO2 (COP 1.39 vs. 0.723, at design conditions) and the high level of direct emitted for the air conditioning in the three different configurations normal irradiation (2296 kWh/m2) available for the concentrated solar was estimated. The assessment is carried out using a conversion factor devices (PTCs). The results presented in this study can be generalized as that takes into account the average efficiency of the Saudi power gen- a guideline for the design of solar district cooling systems in similar site eration systems, as reported in the World Energy Council website [50]. locations (with high levels of beam solar radiation and dry climate). Table 11 shows the primary energy consumption (based on tonnes of oil Furthermore, the paper reports an estimation of the primary energy equivalent) for the required power supply and the related CO2 pro- savings and greenhouse gas emission reduction with respect to a con- duction. ventional district cooling system based on compression chillers: the solar district cooling system for 96 detached homes with solar fraction

0.7 allows to reduce primary energy consumption and CO2 emissions by 4. Economic analysis about 500 TOE and 1400 tonnes per year respectively.

The optimization procedure presented in the Section 2 allowed for References sizing the main components assuring the budget cost minimization. An estimation of the total budget costs for the two solar district cooling [1] Morvaj B, Evins R, Carmeliet J. Optimising urban energy systems: simultaneous fi system sizing, operation and district heating network layout. Energy con gurations is reported in Tables 12 and 13. The goal of the present – fi 2016;116:619 36. analysis is not providing a cost evaluation for a speci c application: a [2] Casisi M, Pinamonti P, Reini M. Optimal lay-out and operation of combined heat & number of details (including construction works, design and labor costs, power (CHP) distributed generation systems. Energy 2009;34:2175–83. consumables, maintenance, etc.) should be evaluated for a real business [3] Oppelt T, Urbaneck T, Gross U, Platzer B. Dynamic thermo-hydraulic model of district cooling networks. Appl Therm Eng 2016;102:336–45. plan. The present study aims to compare only the costs that directly [4] Wang H, Wang H, Zhou H, Zhu T. Modeling and optimization for hydraulic per- depend on the size determined by the optimization algorithm. The formance design in multi-source district heating with fluctuating renewables. values reported in Table 11 are calculated according to the unit costs Energy Convers Manage 2018;156:113–29. [5] Rezaie B, Rose MA. District heating and cooling: review of technology and potential shown in Table 2 and refer to budget costs provided by market op- enhancements. Appl Energy 2012;93:2–10. erators and manufacturers. The costs of the district network (Table 13) [6] Ameri M, Besharati Z. Optimal design and operation of district heating and cooling were estimated using published data [48,49,51]. The cost of the aux- networks with CCHP systems in a residential complex. Energy Build fi 2016;110:135–48. iliary chiller is the same for both con gurations, as it is designed to [7] Romanchenko D, Kensby J, Odenberger M, Johnsson F. Thermal energy storage in cover the peak cooling load in the event of a system failure. The re- district heating: centralised storage vs. storage in thermal inertia of buildings. sulting costs for the chiller are similar for the two investigated config- Energy Convers Manage 2018;162:26–38. urations, whilst the costs of the solar fields are strongly different [8] Olsthoorn D, Haghighat F, Mirzaei PA. Integration of storage and renewable energy into district heating systems: a review of modelling and optimization. Sol Energy (+79.8%). This is due to a slight higher unit cost of ETCs (+4%) and 2016;136:49–64. the higher efficiency of the two-stage absorption chiller. The network [9] Elci M, Oliva A, Herkel S, Klein K, Ripka A. Grid-interactivity of a solar combined – cost accounts for about 15% of the global investment for the case PTC- heat and power district heating system. Energy Procedia 2015;70:560 7. [10] Schmidt T, Mangold D, Müller-Steinhagen H. Central solar heating plants with 2sABS. seasonal storage in Germany. Sol Energy 2004;76:165–74. Such results demonstrate that the solar cooling system based on [11] Powell KM, Cole WJ, Ekarika UF, Edgar TF. Optimal chiller loading in a district parabolic troughs and two-stage absorption chiller is attractive and cooling system with thermal energy storage. Energy 2013;50:445–53. [12] Gang W, Wang S, Xiao F, Gao D. Performance assessment of district cooling system promising for applications in climate conditions similar to Riyadh. coupled with different energy technologies in subtropical area. Energy Procedia Since PTCs can exploit only the beam radiation component, this system 2015;75:1235–41. is expected to perform well in sites with low humidity levels. [13] Liew PY, Theo WL, Wan Alwi SR, Lim JS, Manam ZA, Klemeš JJ, et al. Total Site Heat Integration planning and design for industrial, urban and renewable systems. Renew Sustain Energy Rev 2017;68:964–85. [14] Ozgener L. Coefficient of performance (COP) analysis of geothermal district heating systems (GDHSs): Salihli GDHS case study. Renew Sustain Energy Rev

383 G. Franchini et al. Energy Conversion and Management 166 (2018) 372–384

2012;16:1330–4. [33] Florides GA, Kalogirou SA, Tassou SA, Wrobel LC. Modelling, simulation and [15] Eriksson M, Vamling L. Future use of heat pumps in Swedish district heating sys- warming impact assessment of a domestic-size absorption solar cooling system. tems: short- and long-term impact of policy instruments and planned investments. Appl Therm Eng 2002;22:1313–25. Appl Energy 2007;84:1240–57. [34] Assilzadeh F, Kalogirou SA, Ali Y, Sopian K. Simulation and optimization of a LiBr [16] Lund H, Möller B, Mathiesen BV, Dyrelund A. The role of district heating in future solar absorption cooling system with evacuated tube collectors. Renew Energy renewable energy systems. Energy 2010;35:1381–90. 2005;30:1143–59. [17] Trier D. Towards solar district heating with more than 70% solar fraction. Energy [35] Sokhansefat T, Mohammadi D, Kasaeian A, Mahmoudi AR. Simulation and para- Procedia 2015;70:580–6. metric study of a 5-ton solar absorption cooling system in Tehran. Energy Convers [18] Chasapis D, Drosou V, Papamechael I, Aidonis A, Blanchard R. Monitoring and Manage 2017;148:339–51. operational results of a hybrid solar-biomass heating system. Renew Energy [36] Figueredo GR, Bourouis M, Coronas A. Thermodynamic modelling of a two-stage 2008;33:1759–67. absorption chiller driven at two-temperature levels. Appl Therm Eng [19] Carpaneto E, Lazzeroni P, Repetto M. Optimal integration of solar energy in a 2008;28:211–7. district heating network. Renew Energy 2015;75:714–21. [37] Mazloumi M, Naghashzadegan M, Javaherdeh K. Simulation of solar lithium bro- [20] Zeng J, Han J, Zhang G. Diameter optimization of district heating and cooling mide–water absorption cooling system with parabolic trough collector. Energy piping network based on hourly load. Appl Therm Eng 2016;107:750–7. Convers Manage 2008;49:2820–32. [21] Sameti M, Haghighat F. Optimization approaches in district heating and cooling [38] El Fadar A, Mimet A, Pérez-García M. Modelling and performance study of a con- thermal network. Energy Build 2017;140:121–30. tinuous adsorption refrigeration system driven by parabolic trough solar collector. [22] van der Heijde B, Fuchs M, Tugores CR, Schweiger G, Sartor K, Basciotti D, et al. Sol Energy 2009;83:850–61. Dynamic equation-based thermo-hydraulic pipe model for district heating and [39] Brumana G, Franchini G. Solar-powered air conditioning for buildings in hot cli- cooling systems. Energy Convers Manage 2017;151:158–69. mates: desiccant evaporative cooling vs. absorption chiller-based systems. Energy [23] Chan ALS, Hanby VI, Chow TT. Optimization of distribution piping network in Procedia 2016;101:288–96. district cooling system using genetic algorithm with local search. Energy Convers [40] Nurzia G, Franchini G, Perdichizzi A. Combined solar heating and cooling systems: Manage 2007;48(10):2622–9. simulation and design optimization, ASME Paper ES2008-54127, Jacksonville, FL, [24] Chorowski M, Rogala Z, Pyrka P. System options for cooling of buildings making use USA: ASME International Solar Energy Division; August 10–14, 2008. of district heating heat. Int J Refrig 2016;70:183–95. [41] Hang Y, Du L, Qu M, Peeta S. Multi-objective optimization of integrated solar ab- [25] Arabkoohsar A, Andresen GB. Supporting district heating and cooling networks sorption cooling and heating systems for medium-sized oﬃce buildings. Renew with a bifunctional solar assisted absorption chiller. Energy Convers Manage Energy 2013;52:67–78. 2017;148:184–96. [42] Calise F, d’Accadia MD, Vanoli L. Thermoeconomic optimization of solar heating [26] Lake A, Rezaie B, Beyerlein S. Review of district heating and cooling systems for a and cooling systems. Energy Convers Manage 2011;52(2):1562–73. sustainable future. Renew Sustain Energy Rev 2017;67:417–25. [43] Kirgat GS, Surde AN. Review of hooke and jeeves direct search solution method [27] Horn P, Hauer S, Judex F, Kreulitsch D, Selke T. Solar hybrid heating & cooling analysis applicable to mechanical design engineering. Int J Innov Eng Res Technol systems on district level-The Austrian Project CiQuSo. Energy Procedia (IJIERT) 2014;1(2):1–14. 2016;91:980–8. [44] Remund J, et al. Meteonorm handbook, part II: theory. Bern, Switzerland: [28] Gang W, Wang S, Xiao F, Gao D. District cooling systems: technology integration, Meteotest; 2000. system optimization, challenges and opportunities for applications. Renew Sustain [45] Halldor K, Bøhm B. Advanced and traditional pipe systems: optimum design of Energy Rev 2016;53:253–64. distribution and service pipes. Euroheat and Power 3.IV/2006; 2006: 34–42. [29] Marugán-Cruz C, Sánchez-Delgado S, Rodríguez-Sánchez MR, Venegas M, Santana [46] ASHRAE. Fundamentals Handbook; edition S. I. American society of heating, re- D. District cooling network connected to a solar power tower. Appl Therm Eng frigerating and air-conditioning engineers. Inc; 1997. 33.2. 2015;79:174–83. [47] Schmidt T, Miedaner O. Design Optimization with GenOpt®, Building Energy [30] Perdichizzi A, Barigozzi G, Franchini G, Ravelli S. Performance prediction of a CSP Simulation User News 21 (Sep./Oct. 2000). plant integrated with cooling production. Energy Procedia 2015;75:436–43. [48] Wetter M. Solar district heating guidelines. Solites 2012. [31] Zhai XQ, Qu M, Li Y, Wang RZ. A review for research and new design options of [49] IEA-ETSAP, I. R. E. N. A., Thermal energy storage: Technology brief e17; 2013. solar absorption cooling systems. Renew Sustain Energy Rev 2011;15:4416–23. [50] https://www.worldenergy.org/data/eﬃciency-indicators/. [32] Atmaca I, Yigit A. Simulation of solar-powered absorption cooling system. Renew [51] Davies G, Woods P. The potential and cost of district heating network, 2009 Pöyry Energy 2003;28:1277–93. Energy (Oxford).

384